Stages Documentation

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revisionPrevious revision
Next revision		Previous revision
712:chatbot_config [2025/08/22 20:25] – [Assistant Configuration] Fryer, Kathryn712:chatbot_config [2026/07/21 19:34] (current) – [Limitations] Meier, Erich
Line 6: Line 6:
  
 The sections below describe how to set up, train, and maintain the Stages Assistant. The sections below describe how to set up, train, and maintain the Stages Assistant.
- 
  
 ===== Acknowledgement of AI-enabled Features ===== ===== Acknowledgement of AI-enabled Features =====
Line 15: Line 14:
  
 [[ai_acknowledgements|See here for more info]] on this subject. [[ai_acknowledgements|See here for more info]] on this subject.
- 
  
 ===== Views and Workspaces ===== ===== Views and Workspaces =====
  
-Processes in Stages exist in many different versions, workspaces, variants, and process instances. AI technologies work best if the scope of the underlying information is clear and well-defined. Combining different process versions, variants, and instances would result in imprecise or even misleading answers. Therefore, multiple Stages Assistants can be defined, one for each //view// on the process landscape.+Processes in Stages exist in many different workspaces and versions, including variants and specific project instances. AI technologies work best if the scope of the underlying information is clear and well-defined; combining the information from all Stages processes would result in imprecise or even misleading answers.
  
-Possible process views are:+Therefore, you must define one or more //views// of the process landscape that scope the processes considered by the Stages Assistant. 
 + 
 +Examples of a process view might include:
  
   * Standard processes that everyone should follow   * Standard processes that everyone should follow
-  * Process variants, e.g. a development process for normal products, one for safety-critical products, one of cybersecurity-critical products, and one for both safety- and cybersecurity-critical products +  * A process variant – for example, you might have separate views for a development process for safety-critical products and the process without safety requirements 
-  * Processes that are especially tailored for specific programs or projects+  * A process tailored for a specific program or process 
 + 
 +Views consist of a single workspace or a collection of workspaces. For example, if you have separate workspaces for the standard development process and other organizational workspaces, you can group them into a workspace collection to include all of that process content in a single view. 
 + 
 +Note: The Stages Assistant is always trained on the Valid Version of the process content in each workspace. The Valid Version __must be different__ than the Working Version; if the Working Version is set to be the Valid Version, the training will fail. 
 + 
 + 
 +===== Required Permissions ===== 
 + 
 +End users require the following permissions to use the Stages Assistant: 
 + 
 +  * ''Chatbot'': Read 
 +  * ''Workspace'': Read 
 + 
 +To configure the AI assistants, the following permissions are needed: 
 + 
 +  * ''AI Administration'': RCMD 
 +  * ''Workspace'': Read
  
-Views can consist of a single workspace or a collection of workspaces. This way, the standard process view can consist of a workspace containing the standard development process, one or more workspaces containing all organizational processes that are not executed by projects or product teams, and a workspace containing the organizational policies.+In both cases, the permission scope must include the workspaces for the defined view. If the view includes a workspace collection, the permissions must apply to all workspaces within the collection.
  
-Stages Assistants are always trained on the //Valid Version// of the process content in each workspace. This version __must be different__ than the Working Version.+In the Assistant, users will only see the views for which they have read permissions.
  
 ===== Configuring the Stages Assistant ===== ===== Configuring the Stages Assistant =====
  
-Configure the Stages Assistants in the ''Administration > AI > Chatbot'' menu. The main screen shows the currently configured assistant views.+Configure the Stages Assistant in the ''Administration > AI > Chatbot'' menu. The main screen shows the currently configured assistant views.
  
 {{ :712:chatbotconfig-overview.png?direct }} {{ :712:chatbotconfig-overview.png?direct }}
Line 63: Line 80:
  
   * Name of the view   * Name of the view
 +  * Domain of the processes contained in the view, e.g. "Product Development Processes". It is not required and can be left blank, unless the Assistant answers are too generic or misleading.
   * Welcome Message that end users see when they access the view in the Stages Assistant. The message should describe the scope of information the Assistant can answer questions about.   * Welcome Message that end users see when they access the view in the Stages Assistant. The message should describe the scope of information the Assistant can answer questions about.
   * Color used to display user questions in the Assistant, so users can visually distinguish between different views   * Color used to display user questions in the Assistant, so users can visually distinguish between different views
Line 72: Line 90:
  
  
-===== Acronym Handling =====+===== Defining Acronyms =====
  
-The assistant needs to be made aware of company-specific acronyms that are not described in the process models.+Company-specific acronyms that are not described in the process models must be defined for the Stages Assistant using the ''Administration > Acronyms'' section. Managing acronyms requires administrator access and edit permissions for the "Acronyms" domain.
  
-To manage those acronyms, a Stages administrator with edit permission in the new "Acronyms" domain can access the table via ''Administration > Acronyms''.+For each acronym, you can provide both its expansion and a longer multi-language explanation.
  
-For each acronym, its actual expansion and a longer multi-language explanation can be maintained.+For mass updates, you can import and export CSV files using the "..." context menu.
  
-For mass updates, an import and export of CSV files can be accessed via the "''...''" menu.+===== Disable History =====
  
-===== Required Permissions =====+In newer Stages Versions the chat history can be disabled, e.g., to comply with local policies or regulations. Chats and chat messages will not be stored in the database. To disable history, add the following property to conf/config.xml:
  
-To use the AI assistants as end users, the following permissions are necessary:+<code xml> 
 +<property name="ai.chatbot.history.enabled" value="false"/> 
 +</code>
  
-  * ''Chatbot'': Read +===== Limitations =====
-  * ''Workspace'': Read+
  
-To configure the AI assistants, the following permissions are necessary:+Large Language Models (LLMs) have a so called context window which has a limited size. To accommodate that limit, Stages never sends the process models to the LLM in their entirety, but only in select segments that are required to answer the user questions. As a consequence, questions that address whole process models, e.g. "List all work products of all processes", currently cannot be answered.
  
-  * ''AI Administration'': RCMD +During the training of the Assistant, all process elements are transformed into a knowledge graph and stored in special databases. This means that the Assistant is trained on the following information:
-  * ''Workspace'': Read +
- +
-For both cases, the permission scope needs to be set to the workspaces the chatbots are trained on. In case workspace collections are being used for chatbot training, all workspaces within all collections need to be covered by the permission scope. +
- +
-===== Limitations =====+
  
-The context window of the used AI Large Language Models (LLMs) is subject to limitations. To accommodate those limits, the process models are never sent to the LLM in their entirety, but only in select segments that are required to answer the user questions. As a consequence, questions that address whole process models, e.g. "List all work products of all processes", currently cannot be answered.+  * Element names and abbreviations 
 +  * Element descriptions 
 +  * Associations between elements, e.g. "Role X is responsible for Activity A1"
  
-During the training of the assistant, all process elements are transformed into a knowledge graph and stored in special databases. The following elements are currently excluded in this transformation:+The following elements are currently excluded in this transformation:
  
-  * landing pages +  * Landing pages 
-  * process start pages +  * Process start pages 
-  * "comment only" associations +  * "Comment only" associations 
-  * descriptions that are managed via "''Management > Descriptions''" (except for the inlined ones) +  * Descriptions managed via "''Management > Descriptions''" (except for those used inline) 
-  * graphics that are embedded in element descriptions +  * Graphics embedded in element descriptions 
-  * custom element attributes +  * Custom element attributes 
-  * compliance mappings +  * Compliance mappings 
-  * user-role associations +  * User-role associations 
-  * project files+  * Project files
  
-It is planned, to include those elements in future revisions of the assistant training scripts, if those improve the quality of the given answers.+Such elements might be included in future revisions of assistant training scripts.
  
 Process interfaces to elements outside of the training view (workspace or workspace collection) are ignored. Process interfaces to elements outside of the training view (workspace or workspace collection) are ignored.
  
-To customize the trainings scripts, please contact the Stages support team or your dedicated Stages consultant.+To customize the training scripts, e.g. to have them adjusted to your specific metamodel, please contact your Stages consultant.